Glioblastoma multiforme (GBM), the most common primary brain tumor in adults, is associated with a dismal prognosis of only 12-15 months despite aggressive surgery, radiation, and chemotherapy. The lack of effective treatment options has made this disease a target for new strategies such as gene therapy. However, the only major Phase III clinical trial of gene therapy, involving the use of conventional replication-defective retrovirus vectors in GBM patients, resulted in disappointingly low and therapeutically inadequate transduction levels on the order of only 0.02%. The inability of standard replication-defective retroviral vectors to achieve effective transduction of tumors in vivo is therefore a major obstacle to gene therapy for gliomas. The use of replication-competent vectors for gene transfer would be more efficient, as each tumor cell that is successfully transduced would itself become a virus-producing cell, sustaining further transduction events even after initial administration. We have previously demonstrated that direct intratumoral injection of murine leukemia virus (MLV)-based replication-competent retrovirus (RCR) vector preparations can achieve tremendously efficient suicide gene transfer in gliomas, with transduction stringently restricted to the actively dividing tumor cells without evidence of significant spread to extratumoral sites, and resulting in significantly prolonged survival upon prodrug administration, without detectable systemic side effects. Here we propose to further improve the efficiency of this approach by engineering alloreactive cytotoxic T lymphocytes (alloCTLs) to become RCR vector producer cells, which can then serve as motile cellular delivery platforms that can penetrate into the tumor mass and facilitate multifocal spread of the replicating vectors. Alloreactive CTL exhibit characteristics that provide unique advantages for this purpose: alloCTLs can move through tissue, can themselves kill tumor upon contact, and can produce cytokines that induce apoptosis or can initiate an endogenous immune response. Within CMS gliomas, an immunosuppressive environment within an immunologically privileged site, their own destruction by the immune system may be circumvented long enough for them to have a beneficial effect, but alloCTLs should be rapidly destroyed upon leakage into the general circulation. We will test their viability, biodistribution, safety, and utility as cellular delivery vehicles to enhance RCR vector spread and therapeutic efficacy, comparing immunodeficient and immunocompetent glioma models in vivo.